PhysioNet Challenge 2016

该数据集包含了2575个正常样本和665个异常样本，用于心音分类评估。此外，该数据集从2000赫兹重新采样至16000赫兹，并采用了四折交叉验证的方法进行评估，报告了准确率（UA）和F1分数。数据集的规模为3220个样本，其中包括2575个正常样本和665个异常样本。所承担的任务是心音分类。

This dataset includes 2,575 normal samples and 665 abnormal samples, intended for heart sound classification evaluation. Additionally, this dataset was resampled from 2000 Hz to 16000 Hz, and evaluated using four-fold cross-validation, with accuracy (UA) and F1-score reported. The total size of the dataset is 3,220 samples, which comprises 2,575 normal samples and 665 abnormal samples. The task undertaken is heart sound classification.

搜集汇总

背景与挑战

背景概述

PhysioNet Challenge 2016数据集包含3,126条心音记录，旨在通过算法分类正常与异常心音。数据集特点包括多样化的记录来源、不同质量的录音以及包含噪声的挑战性环境，适用于开发鲁棒的心音分类算法。